{"id":"abd9a7a1-8bb4-4b78-8311-fe4e60aeb853","arxiv_id":"1908.08303","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A complete gradient Ricci soliton with nonnegative Ricci curvature and a convex potential satisfying weighted integral conditions must be Ricci flat and split a line; concave bounded-Ricci potentials force a non-shrinking soliton with at most one critical point of scalar curvature.","lead":"This paper proves rigidity results for gradient Ricci solitons. Under nonnegative Ricci curvature plus a convex potential with finite weighted integral tails, the soliton is Ricci flat and splits off a line; a concave potential with bounded Ricci rules out shrinking solitons.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2.1's estimate from condition (4) is invalid without a sign or absolute-value hypothesis; the natural |u| repair makes Theorem 2.5 vacuous, so the Ricci-flat/splitting claim is not established as stated.","rationale":"The reader's weakest assumption is correct and is the load-bearing point: Lemma 2.1 is the only bridge from the analytic hypotheses to the geometric conclusions. My pass adds two observations. First, the invalid inequality is the exact spot where ∫ u Δφ² → 0 is inferred; without it, Δu = 0 is unsupported. Second, the standard repair u ≥ 0 or |u| is not neutral: for the affine function produced by the theorem, the |u|-integral diverges, so the theorem would have no nonconstant instances. This does not disprove the theorem; it means the statement is ambiguous about the meaning of (4) and the proof is incomplete under any literal reading. A rigorous version would need to specify (4) as, say, convergence of a symmetric improper integral and then prove Lemma 2.1 with a cancellation or sign argument rather than the present inequality. Since the geometric program is plausible and the second half of the paper is standard, the reader's CONDITIONAL verdict is appropriate; my stress test does not move it.","tokens_in":5331,"tokens_out":25285,"duration_ms":283992,"concrete_test":"On M = N × R with N compact Ricci-flat, take u(y,t) = ct, p = (y₀,0), and the radial cutoff of the paper. Compute (i) ∫_{M\\B(p,r)} d^{-2}|∇u|², (ii) ∫_{M\\B(p,r)} d^{-2}|u|, and (iii) ∫_{B(p,2r)\\B(p,r)} u Δφ_r². Verify that (i) is finite, (ii) diverges logarithmically, and (iii) vanishes only by antisymmetry of u and evenness of the radial cutoff. This distinguishes the two readings of (4) and shows whether the lemma's limit can be proved without a sign hypothesis.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim rests on Lemma 2.1, where condition (4) is used in the displayed chain\n0 ≤ ∫ φ_r² Δu = ∫ u Δφ_r² ≤ (C/r²) ∫_annulus u → 0.\nThis chain is not justified: convexity gives Δu ≥ 0, but u itself need not be nonnegative, and Δφ_r² ≤ C/r² does not control ∫ u Δφ_r² when u changes sign. Without u ≥ 0 or an absolute-value tail estimate, the conclusion Δu = 0 does not follow; consequently ∇²u = 0, the Killing/splitting conclusion, and the Ricci-flatness in Theorem 2.5 are not derived. The Bochner step also contains the false local identity ∫_{B(p,r)} (|∇²u|² + Ric(∇u,∇u)) = 0 from φ² ≡ 1, although this half can be repaired by a global cutoff limit. The sign issue is not merely technical: the natural repair ∫ d^{-2}|u| < ∞ is incompatible with the theorem's own conclusion. On the split manifold N × R that the theorem predicts, an affine u with |∇u| = c > 0 satisfies (3), but ∫ d^{-2}|u| diverges logarithmically in the R-factor. Thus either (4) must be interpreted as a conditional improper integral, in which case the displayed inequality still needs a genuinely new argument, or the repaired statement is vacuous.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies complete gradient Ricci solitons and claims two rigidity results. In Section 2, assuming non-negative Ricci curvature and a non-constant convex potential u satisfying the finite weighted Dirichlet integral (3) and the integral condition (4), Lemma 2.1 asserts that the Hessian of u vanishes. The subsequent results then conclude that the manifold is isometric to a product N × R (Theorem 2.4) and that the soliton is Ricci flat with ∇u a Killing vector field of constant norm (Theorem 2.5). A corollary states a harmonic-function version. In Section 3, the authors prove that a non-constant concave potential with bounded Ricci curvature forces the soliton to be non-shrinking (Theorem 3.1), and under 0 < Ric ≤ K the scalar curvature has at most one critical point (Theorem 3.3). The main tools are cutoff functions, the Bochner formula, and external splitting and scalar-curvature theorems.","tokens_in":5558,"tokens_out":10179,"duration_ms":98123,"significance":"If the proofs were correct, Theorem 2.5 would be a clean rigidity statement: complete gradient Ricci solitons with non-negative Ricci curvature and a non-constant convex potential satisfying (3) and (4) would be Ricci flat and split off a line. The statements are concrete and falsifiable, and the proof strategy is natural. The paper does not rely on fitted parameters or self-citations; the main burden is proof rigor rather than circularity. However, the entire classification in Section 2 rests on Lemma 2.1, whose proof contains a sign-error in the integral estimate and a false local Bochner identity. Because these are load-bearing and the most natural repairs either make the statement vacuous or require genuinely new arguments, the announced results are not established as written. Section 3 contains a related but more local limit-passage gap.","major_comments":[{"comment":"The proof of harmonicity uses the chain 0 ≤ ∫ φ_r² Δu = ∫ u Δφ_r² ≤ (C/r²) ∫ u → 0. This chain is not justified by the stated hypotheses. Convexity only gives Δu ≥ 0; the function u itself may change sign, and ∫ u Δφ_r² cannot be controlled by (C/r²) ∫ u unless u ≥ 0 or (4) is replaced by an absolute-value condition. Since (4) is stated for u without absolute value, the conclusion Δu = 0 does not follow. The natural repair, replacing u by |u| in (4), is incompatible with the theorem's own conclusion: on the predicted splitting N × R, a non-constant affine potential with nonzero constant gradient satisfies the weighted Dirichlet condition (3), but the |u|-version of (4) diverges logarithmically in the R-factor. Additionally, the line 'Since φ_r ≡ 1 in B(p,r), using (5), we get ∫_{B(p,r)} Δu = 0' is false, because (5) contains an annulus term ∫_{B(2r)\\B(p,r)} Δu φ_r² that is not shown to vanish. This gap is load-bearing for Lemma 2.1 and hence for Theorems 2.4 and 2.5.","section":"Lemma 2.1, Eq. (4) and the estimate after Eq. (5)"},{"comment":"The displayed local identity ∫_{B(p,r)} (|∇²u|² + Ric(∇u,∇u)) = ∫_{B(p,r)} (1/2)|∇u|² Δφ_r² = 0 is incorrect. On a ball, ∫_{B(p,r)} Δ|∇u|² equals a boundary flux, not ∫ |∇u|² Δφ_r²; the latter vanishes only because φ_r ≡ 1 on B(p,r), while the former need not vanish. A global cutoff argument could repair this by bounding the ball integral by the annulus integral, so this particular error is local and fixable, but as written the proof of Eq. (7) is not valid.","section":"Lemma 2.1, Bochner step around Eqs. (6)–(7)"},{"comment":"The proof asserts lim_{t0→∞} ∫_0^{t0} ∇²u(X,X) dt ≤ 0 and then treats lim_{t0→∞} λt0 as +∞. Neither limit is proved to exist. The argument can be repaired: writing A(t0) = ∫_0^{t0} ∇²u(X,X) dt, concavity gives A(t0) ≤ 0, and Eq. (11) gives λt0 ≤ C + A(t0) ≤ C for all t0, contradicting λ > 0. But the manuscript does not present this limsup-based reasoning, and as written the conclusion λ ≤ 0 does not strictly follow.","section":"Theorem 3.1, transition from Eq. (11) to Eq. (12)"}],"minor_comments":[{"comment":"The abstract contains grammatical and typographical errors, e.g. 'we have showed' should be 'we have shown', and 'In p articular' has a missing space.","section":"Abstract and Introduction"},{"comment":"The notation M − B(p,r) should be typeset as M \\setminus B(p,r) (or M \\smallsetminus B(p,r)) to avoid confusion with the Minkowski difference.","section":"Throughout"},{"comment":"Reference [13] gives the arXiv identifier as 'math.DG/02111159'; the standard identifier for Perelman's entropy paper is math/0211159. Reference [5] gives the year as 1971; the cited article by Fang, Man, and Zhang appeared in 2008.","section":"References"},{"comment":"The second-variation inequality (9) is quoted without specifying the variation used; the authors should indicate that one takes φ times parallel unit normal fields along the geodesic, so that the Ricci term emerges after summing.","section":"Equation (9)"}],"recommendation":"reject","confidential_remarks":"The reader's report and my own assessment agree on the central problem: the proof of Lemma 2.1 is invalid as written, and the natural repairs (u ≥ 0 or |u| in condition (4)) either make the statement vacuous or require substantive new estimates. This is not a case of a merely cosmetic fix, and it affects the main theorems of the paper. The authors may still be able to prove a related result with additional hypotheses, but that would be a substantially new manuscript."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read this before you cite the first theorem. The interesting part is the attempt to force a gradient Ricci soliton with convex potential and nonnegative Ricci to split via weighted Dirichlet and tail conditions. I don't know a prior statement of exactly that combination, so the question is worth thinking about. But the proof of the key lemma (2.1) is broken in a way that matters.\n\nThe first misstep is right after (5): since phi^2 = 1 in B(p,r), the text concludes that the integral of Delta u over B(p,r) is zero. That doesn't follow; the annulus contribution is dropped. The later estimate is more serious: 0 <= integral phi^2 Delta u = integral u Delta phi^2 <= (C/r^2) integral u. The left side is nonnegative because Delta u >= 0, but convexity does not make u nonnegative, so the last inequality has no basis. To make it valid you would need integral |u| or u >= 0, and neither is in the assumptions.\n\nIf you try the natural repair, you run into a worse problem. On the manifold the theorem predicts, N x R, a nonconstant affine u has |grad u| = c > 0, satisfies (3), but integral d^{-2}|u| diverges logarithmically in the R-factor. So with the absolute-value version of (4), the theorem would be vacuous. Assuming u >= 0 has the same issue. This is not a line-by-line typo; the statement as written is not supported, and the obvious fixes don't give a useful theorem.\n\nThe second half is in better shape. Theorem 3.1 uses the standard second-variation inequality along a ray and concavity to rule out lambda > 0. It needs a little limit bookkeeping, but the argument is sound. Theorem 3.3 is an immediate consequence after coupling with the cited lemma on steady solitons. The references are honest and there's no circularity.\n\nWho gets value: specialists in Ricci solitons who want the concave-potential observation. Nobody should rely on the convex-potential splitting theorem until the lemma is repaired or restated. As it stands, I would desk reject; the report can be short. If an editor wants to be generous, a referee could document why Lemma 2.1 fails, but this is not a citable proof.","headline":"The concave-potential half is sound; the convex-potential rigidity theorem is not established because Lemma 2.1's integral estimate needs a sign or absolute value that the assumptions don't supply, and the natural repairs make the statement vacuous.","tokens_in":6090,"tokens_out":14810,"would_cite":false,"duration_ms":160848,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":false},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53C20","53C21","53C44"],"pacs":[],"model":"deepseek-v4-flash","headline":"Complete gradient Ricci solitons with convex potentials are Ricci flat and split a line.","keywords":["Ricci soliton","gradient Ricci soliton","convex potential","concave potential","Ricci flat","splitting theorem","scalar curvature","Killing vector field"],"falsifier":"Compute the boundary term in the proof of Lemma 2.1 on a complete manifold with non-negative Ricci curvature for a non-constant convex $u$ that satisfies (3) and (4) but is negative somewhere outside every ball. The paper's estimate $\\int u\\Delta\\phi_r^2\\le (C/r^2)\\int u\\to 0$ is only valid when $u\\ge 0$ on the annulus; if the direct evaluation of $\\lim_{r\\to\\infty}\\int_{M\\setminus B(p,r)}u\\,\\Delta\\phi_r^2$ gives a nonzero value while (3) and (4) hold, then $\\Delta u\\ne 0$, Lemma 2.1 is false, and Theorem 2.5 loses its proof.","tokens_in":5084,"feed_emoji":"📐","tokens_out":17120,"duration_ms":162151,"temperature":0.7,"pith_summary":"This paper tries to prove a rigidity statement for Ricci solitons whose potential function is convex. It shows that a complete gradient Ricci soliton with non-negative Ricci curvature, a non-constant convex potential with finite weighted Dirichlet energy, and a finite weighted tail integral must be Ricci flat; the potential's gradient is then a Killing field of constant norm, and the manifold splits isometrically as $N\\times \\mathbb{R}$. It further shows that a non-constant concave potential with bounded Ricci curvature forces the soliton to be non-shrinking, and that with positive Ricci curvature the scalar curvature has at most one critical point. The result matters because it extends classical splitting and rigidity theorems from affine or harmonic functions to a natural class of convex potentials.","feed_headline":"Convex potential forces a Ricci soliton to be Ricci flat and split","feed_subtitle":"When a convex potential has finite weighted energy, the soliton is steady and splits as N×R.","key_machinery":"The engine is the interaction of convexity with cutoff functions and integration by parts. A convex function is subharmonic ($\\Delta u\\ge 0$), and the cutoff functions $\\phi_r$ used in the proof satisfy $|\\nabla\\phi_r|^2\\le C/r^2$ and $\\Delta\\phi_r^2\\le C/r^2$; these bounds, together with the finite weighted tail conditions, are meant to make the boundary term $\\int u\\,\\Delta\\phi_r^2$ vanish as $r\\to\\infty$, leaving $\\Delta u=0$. The pointwise identity $\\tfrac12\\Delta|\\nabla u|^2=|\\nabla^2 u|^2+\\mathrm{Ric}(\\nabla u,\\nabla u)$, valid once $u$ is harmonic, then forces both the Hessian and the Ricci term to vanish under non-negative Ricci curvature. For the concave-potential half, the main tool is the second variation of arc length along a ray, which bounds the integrated Ricci curvature and rules out $\\lambda>0$.","core_discovery":"The central claim is that if $(M,g,u)$ is a complete gradient Ricci soliton with non-negative Ricci curvature, satisfying $\\nabla^2 u+\\mathrm{Ric}=\\lambda g$, and the non-constant convex potential $u$ obeys the weighted Dirichlet condition $\\int_{M\\setminus B(p,r)}d(x,p)^{-2}|\\nabla u|^2<\\infty$ and the weighted tail condition $\\int_{M\\setminus B(p,r)}d(x,p)^{-2}u<\\infty$, then $u$ is affine, its Hessian vanishes, $\\nabla u$ is a Killing vector field with constant norm, the soliton constant is $\\lambda=0$, and the Ricci curvature is zero. The splitting theorem then gives an isometry $M\\cong N\\times \\mathbb{R}$ with $N$ totally geodesic. The paper also claims that a complete gradient soliton with non-constant concave potential and bounded Ricci curvature must satisfy $\\lambda\\le 0$, and if the Ricci curvature is positive and $\\lambda\\ge 0$, the scalar curvature has at most one critical point.","pith_inferences":["Going beyond the paper, the failure of the sign hypothesis suggests a concrete repair: replace the signed tail condition by $\\int_{M\\setminus B(p,r)}d(x,p)^{-2}|u|<\\infty$, or explicitly assume $u\\ge 0$; with that hypothesis the integration-by-parts argument in Lemma 2.1 becomes valid and the rest of Theorem 2.5 follows as written.","The method also suggests a finite-energy splitting principle: non-negative Ricci curvature plus a convex function with finite weighted Dirichlet energy may split off a line whenever the function is affine in a weak sense, connecting the result to broader rigidity theorems for harmonic functions.","Since the proof only uses pointwise upper bounds on Ricci along a ray, the concave-potential theorem might hold under a weaker one-sided bound on Ricci rather than the global boundedness assumed here, though the paper does not pursue that generalization."],"forward_implications":["With the convex hypotheses, the soliton constant must be $\\lambda=0$: the soliton is steady, never shrinking or expanding.","The manifold is isometric to $N\\times \\mathbb{R}$, so it contains a line and the level sets of $u$ are totally geodesic.","The gradient $\\nabla u$ is a Killing vector field of constant norm, so the potential is an affine coordinate on the $\\mathbb{R}$ factor.","A harmonic function with finite weighted Dirichlet integral on such a gradient soliton already forces Ricci flatness, as stated in Corollary 2.5.1.","Under a concave potential with bounded Ricci curvature the soliton is non-shrinking; with positive Ricci and $\\lambda\\ge 0$ it is steady and its scalar curvature has at most one critical point."],"supporting_citations":[{"why":"Supplies the pointwise identity relating $\\frac12\\Delta|\\nabla u|^2$, $|\\nabla^2 u|^2$, $\\mathrm{Ric}(\\nabla u,\\nabla u)$, and $\\Delta u$, from which vanishing Hessian and Ricci terms are extracted.","marker":"[1]"},{"why":"Provides the cutoff functions $\\phi_r$ whose gradient and Laplacian bounds drive the boundary terms to zero.","marker":"[3]"},{"why":"Establishes that smooth convex functions are subharmonic, giving the sign $\\Delta u\\ge 0$ that starts the proof of Lemma 2.1.","marker":"[6]"},{"why":"Gives the result that a steady gradient Ricci soliton with positive Ricci curvature has at most one critical point of scalar curvature.","marker":"[7]"},{"why":"The splitting theorem that turns vanishing Hessian into an isometric product $N\\times\\mathbb{R}$.","marker":"[9]"},{"why":"Equates vanishing Hessian with affinity and with $\\nabla u$ being a Killing field of constant norm.","marker":"[14]"},{"why":"Justifies that a non-constant convex function forces the manifold to be non-compact.","marker":"[16]"}],"fun_headline_variants":["Convex potential forces Ricci soliton to be flat and split","Weighted convex potential makes soliton split as N×R","Gradient soliton with convex potential splits off a line","Convex potential in Ricci soliton yields flatness and splitting","Finite-energy convex potential turns soliton Ricci flat and split"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the potential $u$ is non-negative (or at least has controlled absolute value near infinity), because only under that sign condition does the weighted tail integral (4) force the boundary term to vanish and yield $\\Delta u=0$, whereas the paper assumes only the weaker signed integral $\\int u\\,d^{-2}<\\infty$.","fun_headline_variants_meta":{"raw":{"variants":["Convex potential forces Ricci soliton to be flat and split","Weighted convex potential makes soliton split as N×R","Gradient soliton with convex potential splits off a line","Convex potential in Ricci soliton yields flatness and splitting","Finite-energy convex potential turns soliton Ricci flat and split"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000532,"raw_usage":{"total_tokens":2506,"prompt_tokens":834,"completion_tokens":1672,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":450,"completion_tokens_details":{"reasoning_tokens":1587}},"tokens_in":450,"tokens_out":1672,"duration_ms":13234,"temperature":1.0,"reasoning_tokens":1587,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:44:33.378717+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the boundary term in the proof of Lemma 2.1 on a complete manifold with non-negative Ricci curvature for a non-constant convex $u$ that satisfies (3) and (4) but is negative somewhere outside every ball. The paper's estimate $\\int u\\Delta\\phi_r^2\\le (C/r^2)\\int u\\to 0$ is only valid when $u\\ge 0$ on the annulus; if the direct evaluation of $\\lim_{r\\to\\infty}\\int_{M\\setminus B(p,r)}u\\,\\Delta\\phi_r^2$ gives a nonzero value while (3) and (4) hold, then $\\Delta u\\ne 0$, Lemma 2.1 is false, and Theorem 2.5 loses its proof.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the pointwise identity relating $\\frac12\\Delta|\\nabla u|^2$, $|\\nabla^2 u|^2$, $\\mathrm{Ric}(\\nabla u,\\nabla u)$, and $\\Delta u$, from which vanishing Hessian and Ricci terms are extracted."},{"cited_title":"and Colding, T","cited_arxiv_id":null,"evidence_quote":"Provides the cutoff functions $\\phi_r$ whose gradient and Laplacian bounds drive the boundary terms to zero."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes that smooth convex functions are subharmonic, giving the sign $\\Delta u\\ge 0$ that starts the proof of Lemma 2.1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the result that a steady gradient Ricci soliton with positive Ricci curvature has at most one critical point of scalar curvature."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The splitting theorem that turns vanishing Hessian into an isometric product $N\\times\\mathbb{R}$."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Equates vanishing Hessian with affinity and with $\\nabla u$ being a Killing field of constant norm."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Justifies that a non-constant convex function forces the manifold to be non-compact."}],"review_version":1}